Wherever you look, it seems that someone is talking up artificial intelligence (AI), or worrying about the threats that we face as a result of its introduction. Everywhere, however, there are different opinions, and no real picture of any kind of overall view. Silicon chip technology designer ARM saw the same problem, and the company commissioned research to explore views on AI.
The findings of the survey make for interesting reading.
The research was global, covering almost 4,000 participants in the US, Europe (Sweden, Germany and the UK) and Asia (China, Japan, Taiwan and South Korea). It is, however, important to note that those who said that they did not know anything about AI were screened out, as was anyone who said that AI would ‘never’ have an impact on human lives. This means that the findings are skewed towards the view that AI will have an impact—which helps if you want to know what impact people think it will have, but not if you want to make an assessment of views on the size of that impact in the global population.
This, then, is very much a survey among those who know about AI, and believe in its potential. Even with these caveats, however, the headlines are striking.
Almost two thirds, 61%, think that AI will make society better than today, and a massive 92% think that we will see an impact in our daily lives in the next ten years. Over one third, 36%, said that AI was already having a noticeable impact. This may be because of Apple’s Siri and Amazon’s Alexa, both AI-driven applications, and indeed, 77% were aware that Siri was powered by AI.
The three most appealing uses of AI were in healthcare, to diagnose illnesses earlier, in traffic controls to cut congestion, and as personal assistants, although autonomous cars came a close fourth. However, these categories do not link very well to the areas that respondents felt that AI could do a better job than humans: heavy construction, package delivery and driving public transport.
Levels of excitement and optimism about AI were balanced out by similar numbers of respondents who were anxious about it. This is perhaps unsurprising, because the precise changes that will result are unclear. It may also, however, be linked to concerns about security.
Security and other concerns
The risks of hacking and data loss were highlighted by large numbers of respondents, and 85% were concerned about security issues with AI. Links to cloud were a particular issue, with the majority saying that they did not think that sensitive data should be sent to the cloud. The tech industry cannot afford to ignore this issue, as lack of trust could be a serious problem preventing widespread uptake.
Another issue of concern around the world was the impact of AI on jobs. Opinion was divided, with around one third recognising the benefits of having dull or dangerous tasks done by machines, and one third feeling that the changes in jobs that would result from AI were the biggest challenge facing us. Interestingly, respondents generally felt that people would remain able to do a better job than machines. This was even true in sectors like heavy construction, identified as the most likely target for AI introduction. In reality, it seems likely that a ‘mixed economy’ will emerge, with humans and machines working in partnership to augment the abilities of both.
While far fewer people than might be expected were bothered about robots taking over the world, there were some interesting regional differences in concerns. In Europe and the US, respondents worried about the reliability of AI devices. In Asia, the worry was that machines would become more intelligent than humans.
Where will the future take us?
One interesting set of questions concerned trust in AI ‘if it proved to be better than humans at [thing]’. For example, customers were asked whether they would trust self-driving cars if they had fewer accidents than human drivers, and an AI doctor if it was better at diagnosis than a human one. Only around half said that they would prefer the AI system, even if it were better at its job.
Respondents seemed, therefore, to value a ‘human’ input into situations. This, in turn, reinforces the view that the future is likely to belong to those who can work with computers and algorithms, forming a partnership that is considerably greater than the sum of its parts.